Tuesday, April 11, 2006

Profiling

Scribe has identified some of The Problems with Profiling. He also contributed a comment to my recent POSIWID post on The True Motive for Identity Cards.

The first problem is that profiling (as currently practised) doesn't work. It produces too many false positives, and too many false negatives.

For example, simple profiling based on a supposed correlation between name and affiliation is going to produce a lot of anomalies.
  • Richard Rees doesn't have an islamic name.
  • Sharif Abdel Gawad (the Greek Armenian Christian recently selected as a BNP candidate) apparently does have an islamic name. [source: Guardian, April 8th, 2006]
And profiling based on a history of contact with known evil-doers doesn't work for new emerging clusters of evil-doers.

But of course, the problem isn't with profiling as such - it is with stupid and unimaginative and counterproductive profiling. Aha, so the solution is to have more extensive and deeper profiling?

But this just produces a deeper problem. For me, the most interesting aspect of Foucault's account of the Panopticon was not the impact on the prisoners, but the impact on the prison warders - and by extension on the society that employs them. And the more so-called intelligence goes into the Machine, the less intelligence is deployed by the real human beings with "intelligence" in their job titles.

Profiling is essentially an anthropological act - and requires all the intellectual caution and self-awareness that Bateson championed - first as an anthropologist, and second as a systems thinker. Steps to an Ecology of Mind should be required reading for policemen and spies. On second thoughts, maybe that's not such a good idea ...

Technorati Tags: